Comparison · Free and open-source tools
Neurones 3D vs MeshLab
Looking for a MeshLab alternative? Neurones 3D is free, offline point cloud and Gaussian splatting software for Windows: the free viewer runs every process, from AI classification to dimensioned floor plans, and a one-time paid edition saves the results.
MeshLab is an open-source system from ISTI-CNR for processing and editing large unstructured 3D triangular meshes, with point cloud to surface reconstruction, alignment and mesh repair.
By Dr. Florent Poux, maker of Neurones 3D · MeshLab's facts from Visual Computing Lab, ISTI-CNR (open source)'s own documentation, checked
The short answer
Choose Neurones 3D when you start from survey point clouds and need answers from them: classes, floor plans, terrain, contours, change with M3C2 and a report, offline on a licence bought once. Choose MeshLab when the mesh itself is the product: cleaning, repairing, simplifying and aligning scanned meshes for 3D scanning or 3D printing, on any operating system, at no cost, with PyMeshLab for batch jobs.
At a glance
| Neurones 3D | MeshLab | |
|---|---|---|
| Runs on | Windows 10 and 11 (64-bit) | Windows, macOS and Linux |
| Where it runs | desktop | desktop |
| Licence | free viewer + one-time paid editions | free and open source (GPL 3.0) |
| Price | Free viewer; paid editions bought once, no subscription | Free source |
| Free use | The free viewer opens, views and runs every process; writing results out needs a paid edition | Entirely free |
Feature by feature
"Not stated" means we could not find it in Visual Computing Lab, ISTI-CNR (open source)'s documentation, not that it is missing. Each of MeshLab's cells links to where we read it.
| Capability | Neurones 3D | MeshLab |
|---|---|---|
| Runs fully offline | YesNo account, no internet connection, no telemetry | YesDesktop application source |
| One-time licence available | YesBought once; no subscription | YesFree under GPL 3.0; nothing to buy source |
| Opens LAS/LAZ and E57 | YesLAS, LAZ, COPC, E57, PLY, PCD and more; a 127.8 M point E57 opens in about 22 s | PartlyE57 import and export; LAS and LAZ are not in the documented format list source |
| Scan registration | PartlyCloud-to-cloud alignment (ICP and graph matching, no initial placement needed), RMS shown with the inlier share | YesMesh alignment, ICP and a Global Registration plug-in (OpenGR) source |
| AI point cloud classification | YesFour trained models ship in every build: indoor, aerial LiDAR, powerline, streetscape; a confidence per point | Not stated |
| Segmentation | YesUnsupervised segmentation, primitive decomposition, components, rooms | Not stated |
| Automatic floor plans | YesAutomatic, dimensioned, with sections and elevations | Not stated |
| DXF / CAD drawings | YesDXF R2000, SVG and PDF; no DWG | PartlyDXF export of meshes source |
| DTM / DSM | YesDTM, DSM, CHM, slope, aspect, hillshade on one grid | Not stated |
| Contour lines | YesContours that cannot cross | Not stated |
| Change detection | YesM3C2 with a 95 % level of detection; cloud-to-cloud and cloud-to-BIM deviation | Not stated |
| Meshing | YesPoisson and clean meshing; the clean mesh is closed, not edge-manifold | YesPoisson and screened Poisson reconstruction, simplification, hole filling, mesh booleans source |
| Gaussian splats | YesOpens PLY and SPZ splats, converts them to points, meshes them; does not train splats | Not stated |
| Reports | YesSelf-contained HTML reports with provenance | Not stated |
| Scripting / SDK | YesEngine & SDK edition: Python package and headless command line | YesPyMeshLab, a Python library for MeshLab filters source |
Where Neurones 3D is stronger
- Opens LAS, LAZ and COPC as well as E57; LAS and LAZ are not in MeshLab’s documented formats.
- Four trained AI classification models for point clouds, with a confidence per point; MeshLab lists none.
- Automatic floor plans, DTM, DSM and contours, none of which MeshLab lists.
- M3C2 change detection with a 95 % level of detection, and self-contained HTML reports, which MeshLab does not list.
Where MeshLab is stronger
- Free and open source, on Windows, macOS and Linux.
- A deeper mesh toolkit: repair, hole filling, simplification, mesh booleans and colour work for 3D printing.
- Mesh alignment with ICP and global registration, where Neurones 3D aligns clouds only.
Who should choose which
Choose Neurones 3D to take a point cloud or a splat capture to a finished deliverable on your own machine: classified LAS, a dimensioned floor plan with sections and elevations in DXF, a DTM with contours, a change map or a report, without an account, an upload or a subscription. Every figure it publishes is measured; see the benchmarks.
Choose MeshLab: 3D scanning and 3D printing users who clean, repair, align and reconstruct meshes and want a free tool with a Python library for batch work.
Questions
Is MeshLab free?
MeshLab: Free. Free use: Entirely free. Neurones 3D’s free viewer opens, views and runs every process; writing results out needs a paid edition bought once.
Does MeshLab work offline?
MeshLab: yes. Desktop application. Neurones 3D: yes. No account, no internet connection, no telemetry.
What is the best alternative to MeshLab?
Choose Neurones 3D when you start from survey point clouds and need answers from them: classes, floor plans, terrain, contours, change with M3C2 and a report, offline on a licence bought once. Choose MeshLab when the mesh itself is the product: cleaning, repairing, simplifying and aligning scanned meshes for 3D scanning or 3D printing, on any operating system, at no cost, with PyMeshLab for batch jobs.
How this comparison was made
Written by the maker of Neurones 3D, so read it as such. MeshLab's column comes only from Visual Computing Lab, ISTI-CNR (open source)'s published documentation and listings, read on 2026-09-24; where we found nothing, the cell says "not stated". Neurones 3D's column comes from its own guides, which give the limits too. If something here is wrong or out of date, write to howto@learngeodata.eu and it will be corrected.
Sources: